AMD-Pro-BenchCommunity

13,812 benchmarks in 32 GPU combinations documented

Qwen3-8B on AMD GPUs

43 results · 6 sources · 54 measurements

Explore all results

Grouped by reported model identity. Exact weights, quantizations and software can differ. How results are reviewed

Documented results

24 selected tests
View
ReportLocalMaxxing		Qwen3-8BQ4_K_XLModel ref. ↗Token speed—
1,051.5 tok/sPrefill · median · unknown
40.8 tok/sGeneration · median · unknown
1 × Ryzen AI Max+ 395Single GPU · Unified memory
Configuration
Memory
Unified memory
OS
Ubuntu 24.04
Engine
llama.cpp
Backend
vulkan
Harness
LocalMaxxing community speed test; exact client where reported
Prompt / output
1177 / 512 tokens
Context
8192
Concurrency
Not reported
Date
2026-08-12
Full record →
View
Reportllama-bench		Qwen3-8BQ4_K_MWeights ↗Token speed——
21 tok/sGeneration · per request
1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11 (10.0.26200)
Engine
llama.cpp
Backend
Vulkan
Harness
llama-bench
Prompt / output
0 / 16 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-08-06
Full record →
View
Reportllama-bench		Qwen3-8BQ4_K_MWeights ↗Token speed—
2,719 tok/sPrefill · per request
—1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11
Engine
llama.cpp
Backend
Vulkan
Harness
llama-bench
Prompt / output
128 / 0 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-08-07
Full record →
View
ReportServing benchmark		Qwen3-8BFP16Model ref. ↗Conflicting evidenceToken speed
27.7 tok/sThroughput · per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Rocky Linux 10.1
Engine
vLLM
Backend
ROCm
Harness
GenAI-Perf (streaming)
Prompt / output
? / ? tokens
Context
16384
Concurrency
1
Date
2026-09-10
Full record →
View
ReportGeneric LLM		Qwen3-8BQ8_0Model ref. ↗Token speed——
68.82 tok/sGeneration · mean · per request
1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Ubuntu 24.04
Engine
llama.cpp
Backend
Vulkan / Mesa RADV
Harness
Not reported
Prompt / output
? / 128 tokens
Context
Not reported
Concurrency
Not reported
Date
2025-11-18
Full record →
View
ReportGeneric LLM		Qwen3-8BQuant not reportedModel ref. ↗Conflicting evidenceToken speed——
30.2 tok/sGeneration · per request
1 × W7800Single GPU · 48 GB VRAM/GPU
Configuration
Memory
48 GB VRAM/GPU
OS
Ubuntu 24.04
Engine
SGLang
Backend
ROCm
Harness
Not reported
Prompt / output
? / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2026-09-05
Full record →
View
ReportLocalMaxxing		Qwen3-8BQ4_K_XLModel ref. ↗Token speed—
1,029.9 tok/sPrefill · median · unknown
41 tok/sGeneration · median · unknown
1 × Ryzen AI Max+ 395Single GPU · Unified memory
Configuration
Memory
Unified memory
OS
Ubuntu 24.04
Engine
llama.cpp
Backend
vulkan
Harness
LocalMaxxing community speed test; exact client where reported
Prompt / output
1177 / 512 tokens
Context
8192
Concurrency
Not reported
Date
2026-08-12
Full record →
View
Reportllama-bench		Qwen3-8BQ8_0Weights ↗Token speed——
9.5 tok/sGeneration · per request
1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11 (10.0.26200)
Engine
llama.cpp
Backend
Vulkan
Harness
llama-bench
Prompt / output
0 / 16 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-08-06
Full record →
View
Reportllama-bench		Qwen3-8BQ4_K_MWeights ↗Token speed—
2,866 tok/sPrefill · per request
—1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11
Engine
llama.cpp
Backend
Vulkan
Harness
llama-bench
Prompt / output
512 / 0 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-08-07
Full record →
View
ReportServing benchmark		Qwen3-8BFP16Model ref. ↗Conflicting evidenceToken speed
104 tok/sThroughput · aggregate
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Rocky Linux 10.1
Engine
vLLM
Backend
ROCm
Harness
GenAI-Perf (streaming)
Prompt / output
? / ? tokens
Context
16384
Concurrency
4
Date
2026-09-10
Full record →
View
ReportGeneric LLM		Qwen3-8BQ8_0Model ref. ↗Token speed——
62.43 tok/sGeneration · mean · per request
1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Ubuntu 24.04
Engine
llama.cpp
Backend
HIP / ROCm 7.1
Harness
Not reported
Prompt / output
? / 128 tokens
Context
Not reported
Concurrency
Not reported
Date
2025-11-18
Full record →
View
ReportLocalMaxxing		Qwen3-8BQ4_1Model ref. ↗Token speed——
39.29 tok/sGeneration · unknown
1 × Ryzen AI Max+ 395Single GPU · Unified memory
Configuration
Memory
Unified memory
OS
Not reported
Engine
llama.cpp
Backend
rocm
Harness
LocalMaxxing community speed test; exact client where reported
Prompt / output
44 / 256 tokens
Context
4096
Concurrency
Not reported
Date
2026-06-21
Full record →
View
Reportllama-bench		Qwen3-8BQ4_K_MWeights ↗Token speed—
791 tok/sPrefill · per request
—1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11 (10.0.26200)
Engine
llama.cpp
Backend
Vulkan
Harness
llama-bench
Prompt / output
128 / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2026-08-06
Full record →
View
Reportllama-bench		Qwen3-8BQ4_K_MWeights ↗Token speed——
104.3 tok/sGeneration · per request
1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11
Engine
llama.cpp
Backend
Vulkan
Harness
llama-bench
Prompt / output
0 / 128 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-08-07
Full record →
View
ReportServing benchmark		Qwen3-8BFP16Model ref. ↗Conflicting evidenceToken speed
157 tok/sThroughput · aggregate
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Rocky Linux 10.1
Engine
vLLM
Backend
ROCm
Harness
GenAI-Perf (streaming)
Prompt / output
? / ? tokens
Context
16384
Concurrency
8
Date
2026-09-10
Full record →
View
ReportGeneric LLM		Qwen3-8BQ8_0Model ref. ↗Token speed—
3,375.96 tok/sPrefill · mean · per request
—1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Ubuntu 24.04
Engine
llama.cpp
Backend
Vulkan / Mesa RADV
Harness
Not reported
Prompt / output
512 / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2025-11-18
Full record →
View
Reportllama-bench		Qwen3-8BQ4_K_MWeights ↗Token speed—
4,245 tok/sPrefill · per request
—1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11 (10.0.26200)
Engine
llama.cpp
Backend
ROCm / HIP
Harness
llama-bench
Prompt / output
512 / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2026-08-06
Full record →
View
Reportllama-bench		Qwen3-8BQ4_K_MWeights ↗Token speed—
2,908 tok/sPrefill · per request
—1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11
Engine
llama.cpp
Backend
ROCm / HIP
Harness
llama-bench
Prompt / output
128 / 0 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-08-07
Full record →
View
ReportServing benchmark		Qwen3-8BFP16Model ref. ↗Conflicting evidenceLatency1 concurrent request
92 msTime to first token · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Rocky Linux 10.1
Engine
vLLM
Backend
ROCm
Harness
GenAI-Perf (streaming)
Prompt / output
? / ? tokens
Context
16384
Concurrency
1
Date
2026-09-10
Full record →
View
ReportGeneric LLM		Qwen3-8BQ8_0Model ref. ↗Token speed—
3,910.84 tok/sPrefill · mean · per request
—1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Ubuntu 24.04
Engine
llama.cpp
Backend
HIP / ROCm 7.1
Harness
Not reported
Prompt / output
512 / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2025-11-18
Full record →
View
Reportllama-bench		Qwen3-8BQ4_K_MWeights ↗Token speed——
98.8 tok/sGeneration · per request
1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11 (10.0.26200)
Engine
llama.cpp
Backend
ROCm / HIP
Harness
llama-bench
Prompt / output
0 / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2026-08-06
Full record →
View
Reportllama-bench		Qwen3-8BQ4_K_MWeights ↗Token speed—
4,199 tok/sPrefill · per request
—1 × RX 9070 XTSingle GPU · 16 GB VRAM/GPU
Configuration
Memory
16 GB VRAM/GPU
OS
Windows 11
Engine
llama.cpp
Backend
ROCm / HIP
Harness
llama-bench
Prompt / output
512 / 0 tokens
Context
Not reported
Concurrency
Not reported
Date
2026-08-07
Full record →
View
ReportServing benchmark		Qwen3-8BFP16Model ref. ↗Conflicting evidenceLatency1 concurrent request
36 msInter-token latency · Per request
——1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Rocky Linux 10.1
Engine
vLLM
Backend
ROCm
Harness
GenAI-Perf (streaming)
Prompt / output
? / ? tokens
Context
16384
Concurrency
1
Date
2026-09-10
Full record →
View
ReportGeneric LLM		Qwen3-8BQ8_0Model ref. ↗Token speed—
3,214.04 tok/sPrefill · mean · per request
—1 × R9700Single GPU · 32 GB VRAM/GPU
Configuration
Memory
32 GB VRAM/GPU
OS
Ubuntu 24.04
Engine
llama.cpp
Backend
Vulkan / Mesa RADV
Harness
Not reported
Prompt / output
1024 / ? tokens
Context
Not reported
Concurrency
Not reported
Date
2025-11-18
Full record →
Quantizations and software

Quantizations: FP16, Q4_1, Q4_K_M, Q4_K_XL, Q8_0
Engines: llama.cpp, SGLang, vLLM

Explore all 43 results →

This is a selection across sources, not a ranking or a matched experiment. Reading benchmark reports

GPU benchmark pages: W7800 · R9700 · RX 9070 XT · Ryzen AI Max+ 395